Available online at www.sciencedirect.com ScienceDirect Genetically encoded fragment-based discovery 1 Ratmir Derda and Simon Ng This opinion describes recent advances of molecular discovery the fragment of interest. This opinion focuses on chemi- technology dubbed Genetically Encoded Fragment-Based cal post-translational incorporation of fragments Discovery (GE-FBD). GE-FBD starts from a known ligand or (Figure 1b). Alternatively, {FP} libraries can be built from ‘fragment’ that binds to a desired target weakly and often with unnatural amino acids (UAA) that contain fragments as low specificity. Covalent incorporation of fragment into a part of their side-chain (Figure 1c). Selection from {FP} diverse, genetically encoded library of peptides yields a library libraries has been shown to identify ligands in which the of peptide–fragment combinations. Selection from such a fragment covalently linked to peptide segment bind the library has a high likelihood to identify ligands, in which the target with improved affinity and specificity when com- peptides bind to distinct adjacent pockets of the target in pared to the original fragment (Figure 1d). Improvement F FP synergy with the fragment and exhibits enhanced affinity and can be defined as r = KD/ KD > 1 or DDGbind Àlog F specificity when compared to the fragment itself. GE-FBD (r) < 0, where KD is the affinity of the fragment towards could employ fragments that bind non-covalently as well as the target of interest (Figure 1e). Figure 1d summarizes reversible covalent warheads. The key advances in GE-FBD up-to-date outcomes from published GE-FBD reports include (i) synthetic chemistry that enables incorporation of and shows that r = 10–100 can be found across diverse F diverse fragments into both linear and cyclic peptide libraries; fragment with KD ranging from 100 nM to 1 mM [3 ,4– (ii) quantification of multi-step modifications in million-to-billion 6,7 ,8,9 ,10,11 ,12,13 ,14]. GE-FBD was also reported to library members, (iii) and chemical transformations that permit be successful with reversible covalent fragments. incorporation of fragments with concurrent topological change from linear to macrocyclic topologies. Historically, peptides are considered as the suboptimal drug modality due to their poor stability and pharmaco- Address logical properties. However, peptides are increasingly Department of Chemistry, University of Alberta, 11227 Saskatchewan recognized as potential starting point for lead optimiza- Dr., Edmonton, AB T6G 2G2, Canada tion through systematic medicinal chemistry studies. Chemical modification of peptides—cyclization, N-meth- Corresponding author: Derda, Ratmir ([email protected]) 1 ylation, capping of N-terminus and C-terminus and sub- Current address: MSD, Singapore. stitution of L-amino acid with unnatural or D-amino acid— Current Opinion in Chemical Biology 2019, 50:128–137 can yield peptide derivatives with excellent stability and cell permeability that are suitable for pre-clinical and This review comes from a themed issue on Next generation therapeutics ‘investigational new drug’ (IND) studies [15,16]. These efforts have translated to a stapled peptide (ALRN-6924) Edited by Yimon Aye and Paul J Hergenrother that has entered the clinical trial (NCT02264613). The For a complete overview see the Issue and the Editorial encouraging developments show the potential of the Available online 30th April 2019 peptide modality, especially for targeting protein–protein https://doi.org/10.1016/j.cbpa.2019.03.014 interactions that are historically intractable to small-mol- 1367-5931/ã 2019 Elsevier Ltd. All rights reserved. ecule approach [17,18]. Biophysical considerations of GE-FDB GE-FBD offers a unique opportunity to study genetically encoded ligand discovery. Unlike traditional selection from a random library of molecules where the vast major- Introduction ity of ligands exhibit non-detectable binding for the Fragment-based drug design (FBDD) is an important target, members of the {FP} library contain the fragment F process in medicinal chemistry that gives rise to specific, that has a measurable affinity KD towards the target. and selective binding ligands to a protein of interest by Specifically, {FP} library can be demarcated into mem- th linking weak and often promiscuous synthetic molecules bers in which the affinity of i peptide–fragment combi- FPi termed ‘molecular fragments’ (Figure 1a) [1,2]. This nation ( KD) are stronger (DDGbind < 0), similar review focuses on technology that combines the basic (DDGbind 0) or weaker (DDGbind > 0) when compared principles of FBDD with the power of genetically to the fragment F (Figure 1e). Historically, GE-FBD encoded (GE) discovery of peptides or macrocyclic pep- employed multiple rounds of selection and Sanger tide ligands [3 ]. Covalent incorporation of a fragment F sequencing to discover singular ligands with favourable into a GE-library of peptides, denoted as {P} yields a new DDGbind. Use of deep sequencing permitted not only library, denoted as {FP} in which most members contain reducing number of the rounds of selection but also Current Opinion in Chemical Biology 2019, 50:128–137 www.sciencedirect.com Genetically encoded fragment-based discovery Derda and Ng 129 Figure 1 constant random fragment (a) binding site peptides screen Canonical fragments Genetically-encoded fragment-based fragment-based discovery drug design selection / panning (GE-FBD) (FBDD) link fragments (b) post-translational (c) co-translational modification incorposaiton of fragment as the side chain of UAA 20 amino acids Diversification DNA message with repurposed codon 20 + N amino acids (d) increase in affinity (e)Changes in binding of fragment F (f) (g) due to its covalent linking to peptide Pi results of to make a FP combination sequencing i peptide library modified library unmodified peptides chemical post-translational modification Naive Selected Naive Selected (ligation of F to every peptide P) stronger weaker binding binding selection observation in selection experiment selection number of peptides Current Opinion in Chemical Biology General concepts of GE-FBD. (a) canonical FBD versus GE-FBD. (b) Nature relies on PTM to increase functional diversity using 20 amino acids. (c) Incorporation of unnatural amino acids. (d) Affinities of fragments used in GE-FBD and the reported increase in affinity after GE-FBD. (e) Peptide–fragment combination (PF) that have similar, stronger or weaker affinity when compared to original fragment F. Ligands in a {PF} library of FPi modified peptides, thus, exhibit a distribution of binding affinities; GE-FBD strives to identify the lowest KD, but understanding the entire distribution can further illuminate structure–activity relationship (S.A.R). (f–g) Multiset representation of GE-FBD: chemical modification of a peptide library (multiset {P}) creates a {FP} library, in which most peptides bear the fragment and some remain unmodified. Differential enrichment analysis of selections from {P} and {FP} libraries can identify peptide Pi that yield FPi combination that bind with higher affinity than parental fragment [3 ]. changing the analysis approach. Instead of consecutive (FPi,Pi) [3 ]. Including additional controls, such as pan- multi-round selection, the decisions can be made by the ning of {P-F} library on irrelevant target or panning a analysis of multiple parallel instances of selection of both library modified with irrelevant fragment {P-f’} improved modified {FP} and unmodified {P} libraries (Figure 1f) identification of the consensus motifs of peptide frag- [3 ,8,10]. The differential enrichment (DE) analysis ments that bind productively [3 ]. For example, in identify the peptides that are reproducibly enriched in parallel experiments, we selected {FP} library against {FP} but not in {P} libraries. DE analysis calculates the the desired receptor concanavalin A (ConA) and promis- ConA BSA ratio Ri = FPi/Pi and p-value between replicates pi = ttest cuous receptor BSA and defined Ri = FPi/ FPi and www.sciencedirect.com Current Opinion in Chemical Biology 2019, 50:128–137 130 Next generation therapeutics ConA BSA pi = ttest( FPi, FPi) [3 ]. DE analysis flagged libraries for GE-FBD. Chemical post-translational mod- putative ‘non-specific binders’ ligands that exhibited ifications (cPTM) are attractive because one ligation low Ri and high pi exemplifying a statistically insignificant strategy can diversify the same library with many different binding to ConA and BSA. Testing the same {FP} library fragments. Drawback of cPTM is the uncertainty in the against other controls and applying the same DE analysis incorporation of the fragment: as the fragment is not further decreases the likelihood of identifying poly-spe- encoded genetically, the fidelity of fragment incorpo- cific binders. Searching databases of previous phage dis- ration cannot be assessed via sequencing. It is thus crucial play and other selection experiments could serve as to develop methods for quantifying the yield of chemical additional ‘control’. Of course, regardless how many modifications of libraries to ensure the complete conju- controls are performed, it is unrealistic to expect that a gation of the fragments to all members of the library. ligand identified by the screen will bind to only one protein. This specificity should be validated further, for cPTM of N-terminus example, through binding of synthesized ligands to large Oxidation of native N-terminal Ser/Thr residue yields a protein arrays [3]. bio-orthogonal aldehyde handle